Pre-association of polynuclear platinum anticancer agents on a protein, human serum albumin. Implications for drug design

Abstract

The interactions of polynuclear platinum complexes with human serum albumin were studied. The compounds examined were the “non-covalent” analogs of the trinuclear BBR3464 as well as the dinuclear spermidine-bridged compounds differing in only the presence or absence of a central -NH₂-⁺ (BBR3571 and analogs). Thus, closely-related compounds could be compared. Evidence for pre-association, presumably through electrostatic and hydrogen-bonding, was obtained from fluorescence and circular dichroism spectroscopy and Electrospray Ionization Mass Spectrometry (ESI-MS). In the case of those compounds containing Pt-Cl bonds, further reaction took place presumably through displacement by sulfur nucleophiles. The implications for protein pre-association and plasma stability of polynuclear platinum compounds are discussed.

Introduction

Polynuclear platinum complexes (PPCs) containing two or three platinum units linked by alkanediamine or polyamine chains comprise a structurally distinct class of anticancer active platinum compounds, Fig. 1.¹ A distinct feature of these compounds is the presence of a central moiety capable of pre-association or “non-covalent” (hydrogen-bonding and electrostatic) interactions with biomolecular targets. The three global pharmacological factors controlling platinum drug cytotoxicity and antitumor activity are cellular uptake, the frequency and structure of DNA adducts, and the extent of metabolizing interactions. The presence of the central unit (platinum-tetraam(m)ine coordination sphere or the polyamine nitrogen) in polynuclear platinum compounds allows for modulation of these pharmacological factors and may affect drug efficacy.

Fig. 1.

Structures of pairs of trinuclear (BBR3464 I, Ia, Ib) and dinuclear (BBR3571 II, IIa, IIb) studied for pre-association effects on human serum albumin (HSA).

The role of pre-association in dictating the final products of reaction of PPCs with biomolecules has been assisted greatly in the trinuclear case by the synthesis and evaluation of “non-covalent” analogs where the Pt-Cl bonds are displaced by a substitution-inert NH₃ or a “dangling” amine H₂N(CH₂)₆NH₃⁺ (structures Ia or Ib, Fig. 1).^2,3 Similar strategies can be used for polyamine-bridged dinuclear compounds. Further, the synthetic strategies involved in producing linear polyamine-bridged species automatically give rise to a precursor where the central nitrogen is blocked, usually by a carbamate such as t-butyl (Boc).⁴ This pair of compounds then only differs in charge and the presence or absence of a central protonated nitrogen atom. The central nitrogen atom is susceptible to selective attack using, for example, various carbamate or amide blocking groups (structures II, IIa, and IIb, Fig. 1).⁵

The rate of DNA platination as well as the amount of long-range interstrand crosslinks (where the sites of platination are separated by up to 4 intervening base pairs) is affected by charge and hydrogen-bonding capacity of the linker.⁶ For 1,4 interstrand crosslink directional isomers (3′ → 3′ and 5′ → 5′ crosslinks) are formed. The relative proportion of these adducts is dependent on hydrogen-bonding and charge effects within the linker.^7,8 The X-ray crystal structure of a duplex DNA co-crystallized with Ib demonstrated a new mode of DNA ligand binding—a phosphate clamp—where hydrogen-bonding interactions with the phosphate backbone may be arranged in a modular manner.⁹

Intracellular platinum accumulation and cytotoxicity are also affected by charge. In this case, a perhaps counter-intuitive observation is that cell uptake actually increases with charge.^3,10,11 Interestingly, cellular uptake of the spermidine-bridged dinuclear compound II is greater than that of the blocked derivative IIa.¹¹ In these cases enhanced cytotoxicity is always correlated with greater cellular uptake.^11,12 In mechanistic studies with phospholipids, a “non-covalent” interaction of Ia was also observed, suggesting one possible mechanism for membrane transport of these highly charged species.¹³

In this contribution we examine the reactions of polynuclear platinum compounds with human serum albumin (HSA) and show for the first time the presence of a pre-associated or “non-covalent” interaction of a platinum drug on the protein. Reactions in plasma, and especially with HSA, are generally considered responsible for the metabolic deactivation of platinum-based drugs. Understanding the molecular details of such interactions may allow for strategies to manipulate their extent with consequent effects on cellular uptake and even therapeutic index.

Results and discussion

The effects of charge on polynuclear platinum complex binding to human serum albumin (HSA) were examined with two sets of compounds. Firstly, the trinuclear BBR3464 I was compared with its non-covalent analog Ia and secondly, the effect of blocking groups in the central nitrogen of the spermidine-linked dinuclear species was evaluated (II, IIa, IIb). See Fig. 1 for structures. Global changes in protein conformation can be assessed by circular dichroism (CD) and fluorescence spectroscopy.^14,15 The CD spectrum of HSA exhibits two negative bands in the ultraviolet region at 208 and 220 nm characteristic of an α-helical structure of the protein. The fluorescence from excitation of HSA at 298 nm reflects changes in the microenvironment of the tryptophan 214 residue, serving as a tool to obtain information about conformational changes of the protein.¹⁴ A decrease in fluorescence of the lone tryptophan of HSA is interpreted as a conformational change making the amino acid more accessible to water quenching.

Trinuclear platinum complexes

Fluorescence spectroscopy

At t = 0 the relative decrease in fluorescence is concentration dependent for both BBR3464 and the “non-covalent” analog Ia. Over time the relative decrease for BBR3464 is significantly greater than at t = 0 for all concentrations whereas for the “non-covalent” analog there is little or no further perturbation of the system, Fig. 2.

Fig. 2.

Changes in fluorescence of HSA treated with trinuclear platinum compounds. Note no significant change for the “non-covalent’ analog over time.

CD Spectroscopy

The CD spectrum of HSA incubated with PPCs is time and concentration-dependent. Immediately upon mixing 1:1 ratios of Pt complex:HSA the ellipticity of the negative band at 208 nm is increased (to more negative values) and the spectra do not change over a 24 h time period, Fig. 3. At t = 0 the spectra for BBR3464 (3) and Ia (4) are superimposable. In contrast when 10:1 ratios are used, the spectrum corresponding to the BBR3464 reaction (1) changes and the ellipticity is reduced (less negative values). In fact the spectra become similar to those observed for 20-fold excesses of either c-DDP or its trans isomer upon extensive incubation with HSA and diminution of the α-helical content of the protein.¹⁵ Likewise, the binding of the ruthenium antitumor drug trans-[RuCl₄(Im)₂]^- (Im = imidazole) induces considerable concentration-dependent change (up to 15%) of the secondary structure upon binding.¹⁶ These results and others^17,18 are therefore consistent for the concept of bond-formation at the higher concentration. The most likely binding sites in all cases remain the sulfur amino acids and the CD results in general are consistent structural transitions of HSA being dependent on the state of the cysteine Cys34.¹⁹ In contrast to all these cited cases where bond formation is expected to occur, there is little or no change for the “non-covalent” analog (5) at the higher ratio, Fig. 3. Thus, the pre-association is also indicated to actually increase the order of the secondary structure with an enhancement of α-helical content.

Fig. 3.

CD spectra of HSA incubated with compounds I or Ia at different ratios; drug:HSA, 1:1 and 10:1 after 24 h. Note no significant change for the “non-covalent” analog after 24 h. (*) Spectra of I (3) and Ia (4) overlap from 195-250 nm.

ESI-Mass spectrometry

It is reasonable to conclude from both the fluorescence and CD data that the spectral changes induced by Ia, which are invariant over time, represent a “pre-association” or non-covalent binding to the protein. At t = 0 the binding of BBR3464 also represents a significant contribution from the non-covalent “pre-association” and that further changes are due to bond formation. To confirm the presence of this interaction, the compound Ia was incubated with rHSA and the mixture examined by ESI-mass spectrometry, Fig. 4. It is observed in the mass spectrum that a noncovalent interaction between the polynuclear platinum compound Ia and rHSA takes place after a very short incubation period. There is a 1015 Da shift in the deconvoluted spectrum (Fig. 4(B)) obtained after incubation reveals that rHSA which corresponds to a 1:1 complex Ia with rHSA with the presence of one of NO₃^- counter ion from the complex. Mass spectrometry has also been used to directly observe c-DDP binding²⁵ and also Cys34 binding to gold of gold-based antiarthritic drugs.²⁰ In the present case initial studies with BBR3464 caused precipitation and no clean spectra were obtained—further studies are in progress.

Fig. 4.

(A) Nano-ESI-MS of rHSA incubated with Ia. rHSA-ligand bound peaks are designated with (*). (B) Deconvoluted mass spectrum showing a mass shift of 1015 Da, corresponding to the addition of Ia + one NO₃^-.

Dinuclear spermidine-linked compounds

Designed synthesis of polyamine backbones gives a series of exceptionally potent compounds which mimic successfully the biological activity of BBR3464.^4,12,22 The critical feature in the biological activity is linked to the presence of a central charged -NH₂-⁺ on the polyamine linker. The role of pre-association in dinuclear polyamine-linked compounds can be examined by comparing the reactions of the charged species and its uncharged or blocked analog. Interestingly, the relative decrease in fluorescence at 1:1 ratios was significantly less for the blocked derivatives. In this case, however the presence of Pt-Cl bonds does mean that over time platinum-protein bond formation may take place and this is reflected in an enhanced fluorescence decrease over time for all compounds, Fig. 5. It is of interest to note however, that this decrease is not as great as that seen for BBR3464. In agreement with the changes in fluorescence the CD spectrum of HSA bound to BBR3571 after 48 h appears similar to that of BBR3464 at high Pt:HSA (10:1) ratios (data not shown).

Fig. 5.

Changes in fluorescence of HSA treated with dinuclear platinum compounds.

It is generally considered that chemical changes of platinum compounds on HSA lead to deactivation.^14,21 A study of a series of mononuclear and dinuclear derivatives of meso-1,2-bis(4-fluorophenyl)ethylenediamine showed strong binding to HSA by hydrophobic and electrostatic interactions, leading to reduced cellular uptake and decreased potency.²³ A pharmacological principle is that free (non-protein-bound) drug is active, being able to diffuse readily to tissue. Therefore to evaluate and quantitate the binding of PPCs to HSA, BBR3464 and Ia were incubated and the ultrafilterable protein unbound fraction was quantified for selected time points, Fig. 6. Analysis by ICP-OES shows a decrease in Pt content immediately upon mixing, t = 0, for both BBR3464 and Ia. This initial decrease in Pt content represents the non-covalent, “pre-association” of the Pt complex with protein. The Pt content corresponding to BBR3464 decreases with time, representing a displacement of the Pt-Cl bond and a formation of the covalently bound Pt-protein species (where covalent is used to indicate Pt-amino acid residue formation independent of strict contributions of covalent or coordinate bonding). In contrast, ICP-OES analysis of Ia shows a constant Pt concentration from t = 0 through t = 24 h. Due to the presence of the substitution-inert -NH₃ group, the initial pre-association is not followed by a covalent Pt-protein interaction. Similar behavior was observed for Ib.

Fig. 6.

ICP-OES analysis of Pt content in ultrafiltrate of BBR3464 and Ia after incubation with HSA for selected time points.

Conclusions

The results show that the initial reaction of human serum albumin with trinuclear platinum compounds is one of “pre-association”, similar to that observed for DNA and phospholipids. The mass spectral confirmation is the first, to our knowledge shown for platinum antitumor agents. The strength of binding is apparently stronger than that seen for free spermine, spermidine or the cation [Co(NH₃)]₆³⁺, where the CD changes upon incubation were minimal compared to free HSA.²⁴ The “pre-association” observed for I and Ia can be also observed for complexes II—where the presence of a blocking group diminished the initial interaction as a cause of diminished charge and hydrogen-bonding capacity of the central nitrogen. The proposed binding sites for platinum on HSA are the Cys-34 and Met-298, located on the exterior of the protein in sub-domains I and II.^14,15,21,25

There are a number of interesting points to consider based on this understanding. Human serum albumin is a versatile transport protein and receptor for a variety of ligands and drugs. The primary function of HSA is the transport of fatty acids but since it carries a negative 17 charge, it is also responsible for shuttling a wide variety of metal ions, steroids, and a variety of pharmaceuticals. The crystal structure of human serum albumin has allowed the identification of the principal ligand-binding domains located in the hydrophobic cavities of sub-domains IIA and IIIA.^26,27 In principle the receptor sites for the highly-charged cations as represented by PPCs in this study should be distinct from the Cys and Met covalently-binding sites. It may, therefore, eventually be possible to plot the migration of the pre-associated form to the platinated species. Secondly, the identified metabolic products of BBR3464 can be mimicked by reactions with small sulfur nucleophiles such as glutathione and methionine.^28-30 HSA constitutes roughly 60% of the mass of plasma proteins with a concentration of ∼40 mg mL^-1. It is highly likely therefore that HSA binding could contribute to these metabolic deactivation reactions, as suggested also for some similar dinuclear complexes.²³ Finally, as a corollary, the results for Ia and Ib suggest that these non-covalent compounds may “by-pass” the deactivation associated with Pt-S bond formation. This is especially important for Ib. This 8+ compound is taken up into cells in a significantly greater concentration than either BBR3464 or Ia, and as a result has demonstrated in vitro cytotoxicity equivalent to cisplatin. The “non-covalent” compounds have shown a new mode of DNA binding distinct from intercalation and minor-groove binding. The interaction, but not deactivation, on HSA suggests a new and promising profile for a polynuclear platinum drug.

Experimental

All platinum compounds were as prepared previously.^2-5

Sample preparation

Human serum albumin (essentially fatty acid free) was purchased from Aldrich/Sigma. Recombinant human serum albumin (rHSA) (Recombumin®), expressed in Saccharomyces cerevisiae, was a generous gift from Delta Biotechnology Limited (Nottingham, UK). A stock solution of phosphate buffered saline ([phosphate] = 150 mM, pH 7.35 at 37 °C, [NaCl] = 120 mM, [KCl] = 2.7 mM)) (PBS) was prepared and used for all reaction solutions in order to mimic physiological conditions. HSA was incubated with platinum reactions at a constant temperature of 37.5 °C prior to obtaining spectroscopic data, which was acquired at room temperature.

UV-Visible spectroscopy (UV-Vis)

Absorption spectra were recorded on a JASCO V-550 UV-Vis spectrophotometer. HSA concentration was determined from the absorption spectrum, taking the absorbance of a 1 mg mL^-1 solution at 280 nm (λ_max Trp-214) as 0.55.¹⁵

Circular dichroism (CD)

CD spectra were measured on a Jasco J-600 UV dichrometer using a 0.3 mm (open-ended side plate) quartz crystal cuvette. A bulk solution of HSA was prepared at a 38.5 mg mL^-1 (1.12 × 10^-4 M) concentration. HSA/Pt reactions were conducted at a 1:1 and 10:1 (drug:HSA) stoichiometry. Spectral measurements were recorded at t = 0 and selected time intervals.

Fluorescence spectroscopy

Fluorescence measurements were conducted on a Cary Eclipse Fluorimeter with the excitation and emission wavelengths set at 279 and 310 nm, respectively, with a scan rate 120 of nm min^-1. PMT voltage was set manually to 409 V. The reactions of platinum compounds were studied at various reactant ratios, predominantly 1:1 and 10:1 (Pt:HSA). An initial concentration for a 10:1 reaction was prepared for each drug. Each solution was then diluted to the specific concentration needed for the desired ratio. Each reactant was incubated in a shaker bath at 37.5 °C for 45 min prior to reacting. Separate measurements were taken at t = 0 and appropriate time intervals.

Electrospray Ionisation Mass Spectrometry (ESI-MS)

Experiments were conducted on a Waters/Micromass (Manchester, UK) QTOF-2 mass spectrometer. A custom built nano-ESI source was used with a capillary voltage of 2.1 kV and a cone voltage of 36 V. Source block temperature was maintained at 150 °C. ESI-solutions consisted of methanol:water (50:50 v/v) with 0.1% formic acid. The solutions were infused at 0.5 μL min^-1 using a Harvard Apparatus Model 22 syringe pump (Harvard Apparatus, Holliston, MA). Data acquisition and processing were carried out using the MassLynx 4.0 software. Stock solution of AH 44 was 500 μM in water and made fresh before ESI-MS experiments. Aliquots from the stock solution of rHSA, 139 μM, were mixed with the polynuclear platinum compound Ia in a 1:2 molar ratio in water and incubated at 37 °C, for 15 min. Incubated samples were mixed in the ESI solution to give a final rHSA concentration of approximately 40 μM.

Inductively Coupled Plasma-Optical Emission Spectroscopy (ICP-OES)

1.8 mM solutions of each complex were prepared in phosphate buffer and added to 0.6 mM HSA such that the final concentration of complex was 0.9 mM; (3:1; drug:HSA) The mixture was reacted at 37 °C and 100 μL aliquots were removed at t = 0, 0.5, 2, 6, and 24 h. Aliquots were then centrifuged through a Microcon YM-10 10 000 kDa membrane filter at 14 000 g for 30 min to isolate ultrafiltrate. The centrifuge cup was upturned in a second eppendorf and then centrifuged at 1000 g for 3 min to isolate the protein-bound fraction. Samples were digested following published methods. Pt content was recorded at 214 nm and 265 nm, at each time point, using a Varian ICP-OES. Data are reported as percentage of Pt in control vs. Pt per time point.